November 23, 2003
Reaction to Littlejohn
Randy Littlejohn recently wrote a Gamasutra article (free registration required) in which he’s “agitating for the creation of a new kind of interactive experience that is comfortable and compelling for the masses,” namely interactive drama. This post is a reaction to the article, in the process sounding the same horn you’ve heard from me in previous posts, but with a few new comments inspired by the article. As usual I’ll use Façade in most of my talking points even though very few people have gotten the chance to play it, as it’s in its final months of development; again we apologize for that.
These reactions will make the most sense if you’ve already read the article — if you haven’t yet, I encourage you to do so!
Overall I strongly agree with Littlejohn; it’s exciting to hear someone proselytize interactive drama like this.
“What I see is an interactive drama for the masses who have computers, but who are not ‘gamers’.” While I think interactive drama may not necessarily please all ‘gamers’, some of whom may prefer competitive, goal-oriented, action / strategy based experiences, I think interactive drama should still try to do what games do better than any other interactive experiences to date — create rich worlds that offer players a high degree of agency. That is, it would be disappointing if interactive drama became, for example, a lightweight, few-discrete-choice-points kind of experience. Today’s best games make a strong case that continuously interactive, immersive, dynamic and customizable virtual worlds will offer the most satisfying experience for players (and therefore the most lucrative, from a business point-of-view). (In a letter to the editor, Patrick Dugan makes a similar point.)
“Drama is smoke and mirrors – its elements only need to seem real. Bringing the metaphor back to AI, there doesn’t need to be “real” understanding by the machine, or any “real” communication to make interactive drama work. It must only appear to the experiencer that NPCs are capable of real intelligence, understanding, needs, goals, emotion and communication.” I used to think this way. When Michael and I started Façade, we planned to build lots of technology (which we did), but we (especially me) thought that clever design (“smoke and mirrors”) would get us at least half the way. Design is super-important for what works about Façade, and it always will be in any interactive art / entertainment, but we’ve realized that you just hit a brick wall eventually. At a certain point it just becomes too cumbersome to brute-force or fake the kind of complexity we’re talking about here. To really make strides in this area, some real intelligence is truly required. This doesn’t mean human-level intelligence, but it means more than the insect-level intelligence that today’s games have. We need to at least get closer to real animal-level intelligence. (More on this issue in future posts / papers…)
So I guess I’m already somewhat challenging one of Littlejohn’s main points, “I look out there and see that all of the components for a test bed are now available (though dispersed in various computer game development tools and non-entertainment projects).” I think some light-ish-weight initial version of interactive drama (e.g., something Façade-like) could be done with today’s technologies, but I think players will quickly want more sophistication, and it’s going to require some non-trivial (but eminently do-able!) advances in AI technology (integrated with new game design techniques) to get there… (A letter to the editor by Greg Morchower makes a similar point. I firmly agree with Littlejohn’s reply to that letter — “Looking ahead is a good thing, even if blue-sky ideas cannot be achieved immediately. If a new gold mine can be foreseen in the distance, it is wise to begin preparing the road now.” )
Littlejohn envisions a “new kind of production tool would be designed for a dramatist who is not necessarily a programmer. To program well takes years of dedication. To become an excellent composer takes years of dedication. To become a talented animator takes years of work. And to become a writer-dramatist takes years of dedicated work too. It makes sense to let each talent area work at what they do best. It is unrealistic to think that a single person can be a talented AI programmer, a compelling writer AND an insightful dramatist. I’m sure that somewhere such a renaissance person exists, but can we realistically expect one of these rare people at each interactive company?
This new development tool for the non-programmer writer-dramatist would allow for the development of at least characters and interactive story elements. The tool would need to plug into world-creation tools, such as existing level editors.” This does seem very important, and I think will eventually happen… but at first it will probably be necessary to rely on writer/designer/programmers, until such tools are developed (they’ll probably be the ones making the tools). I think the tools will fall out of the new AI technologies mentioned in the previous paragraph. (We’ve had good discussions about the artist-programmer issue on this blog.)
Littlejohn proposes some approaches to architecting an interactive drama system and authoring pathway. I agree with the general gist of his suggestions; the devil’s in the details of course… It should be true that techniques from various related AI technologies could be cherry-picked and adapted for use in building an interactive drama architecture. I think a strength of Façade’s is its integration of a collection of heterogeneous technologies.
Littlejohn discusses Façade towards the end of the paper. He advocates its approach, with some caveats. I’ll try to clear up a few misconceptions. “Missing in Façade is the idea of the characters coming to the stage with actionable “attitude” – the end-result of the characters innate tendencies and past experiences as described by a dramatist in a biography. Instead, I get the feeling that the characters come to the stage with neutral characterization, but are inclined to take certain actions, per the author’s beat goals for that character. This is a subtle difference, but drama is about subtlety.” Littlejohn has only had our papers to rely on for information, so this misperception occurred. Just to clarify — the nature of the behaviors of Grace and Trip (the NPC’s in Façade) are dominated by their backstory. In fact they are almost solely composed of actionable attitude; they have context-specific, non-neutral characterization, to the point where I wish they had a more capability to act in a more general way sometimes.
“Also missing is an authoring interface, which would allow a dramatist who is not a programmer to create dramatic works with the drama world engine they’ve created.” This is very true. Chaim Gingold makes this point strongly in his Masters thesis (which I’ll be reacting to in another blog post in the near future, by the way). While building playable systems is interesting and important, of course it makes tons of sense to get the technology into the hands of other people. Witness the interactive fiction community, with technologies such as Inform. Regarding Façade, understandably I think, our initial goals for the project were just to get a polished interactive drama experience up and running; publicly-releasing the technology could theoretically follow later. Michael has plans to eventually release some of the core languages he developed for the project; we may release some Façade-specific code too, we’re still figuring that out. (A word of warning, we made little or no progress on the non-programmer-as-author issue.) Also, all along the way we’ve been publishing detailed descriptions of the architecture.
“I think that it’s important to create a utilitarian drama world engine instead of worrying about narrative content at the beginning. In short, I think that it’s a mistake to create a drama world engine around a specific narrative, and around specific people.” True, although to be clear, that’s not what happened in Façade; the architecture and its languages are general enough that many kinds of interactive dramas could be created in it, by people other than Michael and me. To whatever extent the Façade technology is released (plus lots of documentation to describe how to use it), that becomes more possible… Again the papers describe the system in enough detail to hopefully make it possible to build one’s own version of the technology. One of our goals (especially Michael’s, as he’s in academia, but I share the goal) is to do research, build systems, and share the results. We will be giving a programming talk at GDC this March to that effect.
One more reaction to a question Littlejohn had about Façade: “Emotional inputs that can be “tuned” in real time is something that I don’t think Façade has.” In Façade, much of the emotional variability has been folded into the narrative variability; that is, the characters in Façade are reasonably emotionally responsive, by way of the narrative itself changing. The narrative state and their emotional state are pretty intertwined. But, that said, the characters can in fact perform the same narrative content in several different ways, depending on their (simple) emotional state. Pulling this off was one of the more time consuming aspects of the project.
But more generally, it’s true that Grace and Trip’s needs and wants don’t have a wide range; in Façade there is no emotional slider to make them start wanting radically different kinds of things. That is a more general approach that you see in the Sims, something I did a lot of in Petz and Babyz, and Michael did in Office Plant #1. We didn’t have time to add that kind of range in Façade, although we wanted to. If we did, we probably would have developed and integrated something very similar to Scott Reilly’s Em system, one of the technologies developed by the CMU Oz project, just as the programming language ABL (better described here), a core piece of Façade, is a descendant of the Oz project Bryan Loyall’s Hap (see the Oz project page for Reilly’s and Loyall’s theses, each only available in postscript. Michael was the last remaining Oz project member, by the way.)
Of the other projects Littlejohn mentions, Cynthia Breazeal’s is the one I look forward to the most (see her book). She’s probably even more crazy and ambitious than we are! (But then there’s Steve Grand, see book…)
The text of this blog entry is licensed under a Creative Commons License.
November 23rd, 2003 at 2:01 pm
My main goal in writing the article was to get people talking, perhaps arguing about the potential of interactive drama. This is a wonderful critique of my critique – exactly the kind of information-packed reaction that I had hoped for. I have great respect for you two. You’re actually doing what I can only write about. I look forward to what you guys do next.
November 24th, 2003 at 4:59 pm
Chris Crawford responds to Littlejohn’s article and readers’ comments on Gamasutra Letters to the Editor.
November 24th, 2003 at 7:45 pm
My main problem with Littlejohn’s article — aside from the freewheeling style of the piece — is that he proposes a drama engine without addressing a market. (I use the term ‘market’ here, ‘audience’ or ‘culture’ might be better, or perhaps ‘program’ in the architectural sense.) You don’t only need an audience, you need to attract artists who will be interested in the form you’ve created. Personally, I’m not interested in a three-act, immersive 3D environment for constructing narratives, but that is a model that many people seem to think will be central to the forthcoming market for interactive drama. Form is not enough, you need something to attach it to, it needs an idiom. And the idioms we currently have for immersive 3D environments are too limited.
November 25th, 2003 at 2:39 am
I think that’s totally cart-before-the-horse-putting. 3D game engines didn’t start happening until people had developed enough 3d games that they had a pretty good idea of what was and wasn’t needed in them. Toolkits are kits of previously chosen tools; the new ground here has to be, uh, forged, by people who have the ability to make and customize their own tools. (Sorry for the mixed metaphor there.) And I think we have a long way to go before the West is no longer Wild.
Another example–the two biggest modern IF development systems, Inform and Tads, were both written as tools for their authors to make their own IF works first, and only as a tool for others to use later on.
And presumably this is the route Façade has gone. I wouldn’t start saying Façade is general enough for making other things until people actually do so; it’s like saying a priori that the Unreal engine is sufficient for implementing a game like Thief or Deus Ex. (Deus Ex was implemented in Unreal… just barely.)
November 25th, 2003 at 4:24 am
Dialogue seems to me the major stumbling block of interactive fiction. We know people primarily through speech; yet it is speech that dooms all computer-generated characters to cardboard cutout-dom. Believable characters must be able to respond with something more than predefined phrases. They must be able to have simple, believable conversations with users. Achieving this, however, probably means passing the Turing Test; which is not a likely prospect, in my opinion.
November 25th, 2003 at 10:20 am
Brandon (whose presentation of the fascinating Dr.K Project I greatly enjoyed at the 1999 NI symposium) says:
[LIttlejohn] proposes a drama engine without addressing a market.
… three-act, immersive 3D environment for constructing narratives …
Form is not enough, you need something to attach it to, it needs an idiom
Yeah, of course we can all speculate but no one knows yet what this will be. Littlejohn alludes to conventional idioms, movies and sitcoms — “This new art form would immerse the experiencer inside a reality very much like what he or she is already familiar with: film and television,” but he doesn’t expand much further. Obviously lots of ideas will have to get built to figure it out. Once it’s better understood maybe we’ll eventually invent a new term or terms for it. Just like with games, predictably there’ll be all sorts of sub-genres… Besides the idea of the Holodeck, The Diamond Age presents a networked, non-AI-based (if I remember correctly) approach (which now seems all the more feasible, now that MMOGs are up-and-running), but the more recent Literary Devices by Richard Powers offers an equally compelling vision, I thought.
Sean writes:
I think that’s totally cart-before-the-horse-putting.
I wouldn’t start saying Façade is general enough for making other things until people actually do so
True; we should qualify that to say “we imagine / intend” the architecture to be usable by others. :-) (Littlejohn’s article possibly implied that we didn’t have that intent at all.)
O writes:
[Characters] must be able to have simple, believable conversations with users. Achieving this, however, probably means passing the Turing Test
Here’s where careful design will save the day, I think. Yes we’ll need lots more natural language processing and discourse management technology than we currently have, to even achieve simple conversational ability. But characters by no means will need to pass a generic Turing Test; interactive dramas can be pretty context-specific, greatly narrowing the domain of what’s required in conversation. Within that domain, they need to be robust and responsive and as seemingly sentient and intelligent as possible, but that domain (by design) can be believably limited, I think. Based on our experience experimenting with this in Facade, I’m reasonably optimistic that it’s possible (with a bigger team of people) to pull off something believable and satisfying to play…
The proof’s in the pudding with all of this speculation, of course. Randy writes:
You’re actually doing what I can only write about.
I want to say that no, it’s not that hard. To any motivated people out there: read all the must-reads, read up on the various approaches being taken in research labs, think about the problem for a while, and you’ll probably come up with an approach that would make some progress in this area. If you’re not a programmer, team up with someone like-minded. If need be, go to grad school. … Brandon independently built the aforementioned Dr.K project; as a Masters student Chaim built the under-development but playable Comic Book Dollhouse. Both are really interesting new approaches to interactive narrative, built by individuals. There are accessible technologies available. It’s do-able, people, you just gotta make it happen.
November 25th, 2003 at 2:10 pm
Sean Barrett wrote: 3D game engines didn’t start happening until people had developed enough 3d games that they had a pretty good idea of what was and wasn’t needed in them.
On the contrary, I think 3D engine creators picked up whatever interesting bits they saw; they grabbed CAD file formats, scene graphs, and texture mapping and threw them all together. We’re only one or two generations away from those “old” engines.
At the same time, we’re at least five years away from the kind of 3D technology that will really facilitate the kinds of interactivity and art I’d like to have. But it is hardware I’m waiting for, not software and toolkits.
Andrew, who quoted Littlejohn: … inside a reality very much like what he or she is already familiar with: film and television.
But who need a new idiom that is just like the old ones? We already have film and television, and they are intractable as interactive forms.
One of the things that is happening — or was happening, things are a bit stale now — with New Media is that people are building and promoting new forms. CAVE environments, Crawford’s Erasmatron, Littlejohn’s proposal. But are people supposed to take these forms and put their own stories in them? Yes, it seems so. But who are the people who are willing to do this? They aren’t the New Media people, because they are interesting in creating their own forms.
At the same time, the people trying to promote their new forms bemoan all the attention that the established game industry gets. Yet another first person shooter, yet another real time strategy game. Given the success of the gaming industry, we should be placing them next to film and television as winning forms. Looking at games, film, and television, what specifically do we want to borrow from them when we create new forms?
A three act structure? Did Zork have a three act structure?
November 25th, 2003 at 3:47 pm
So, I had written a long post, which Mozilla decided to trash before I posted it…let’s see…
To really make strides in this area, some real intelligence is truly required. This doesn’t mean human-level intelligence, but it means more than the insect-level intelligence that today’s games have. We need to at least get closer to real animal-level intelligence.
I agree with this and LittleJohn’s points about examining existing AI techniques as pointers for what capabilites we can incoporate into synthetic characters, automated storytellers / directors, etc. to help create a denser and more believable dramatic environment. For example, I think that one of the big wins in Façade, as Andrew mentions above, is this creative, heterogeneous combination of tools.
One of the main reasons we’ve chosen Haunt 2 as the focus of our work is that it offers a chance to center the gameplay around the “capabilites the AI buys us.” The User is a ghost character in the narrative and can “possess” synthetic agents in the world. Since those agents are well-defined with their own explicit goal structure (i.e. they are Soar agents) the User can actually observe and affect that goal structure, thus altering the agent’s high-level behaviors.
While taking existing AI techqniues and trying to understand how they can fit into interactive drama isn’t the end, finding creative ways of putting them together (e.g. Facade, or Marc Cavazza’s “Friends” project at the U. of Teeside, or the other groups Andrew has posted about) is a great step in the right direction.
November 20th, 2004 at 1:35 pm